strategic and structured approaches in automated …strategic and structured approaches in automated...

Strategic and structured approaches in automated UPLC/HPLC method developmentUPLC/HPLC method development

systemsRudy Sneyers Jeroen Peeters Willy Janssens Luc Van Grieken and Gaby Török

June 9 - 10, 2005ANG annual meetingLatina, October 19 - 20, 2006 1

Rudy Sneyers, Jeroen Peeters, Willy Janssens, Luc Van Grieken and Gaby TörökJohnson & Johnson Pharmaceutical Research & Development

Pharmaceutical Development & Manufacturing Sciences Small Molecules- Method Development

Beerse, Belgium

Purity and Stability-indicating Assay Methods

If no standard components are available, method development can be performed with representative sub-samples to create one single composite sample. For the development of purity- and stability-p p p p y yindicating HPLC methods, the following sub-samples are mainly used:

Raw production batch and starting materials to cover the impurity profile profile Representative forced degradation samples to cover the primary degradation profileStressed drug product formulations (if available)

UPLC

g p ( )

Representative samples for Method

June 9 - 10, 2005ANG annual meeting2Rudy Sneyers J&J PRD

Analytical Development - DS&CC 22

pDevelopment

JNJ27387581_RSV 08-Mar-2006W1_COL1_70_S2 JNJ27387581_RSV 08-Mar-2006W1_COL1_70_S3

Composite and Sub-Samples_

17:18:33_ _ _

-20

100

%

W1_COL1_70_S3 2: Diode Array TIC

1.50e7

W1 COL1 70 S2 2: Diode Array

Excipients17:54:28

100W1_COL1_70_S3 2: Diode Array

TIC1.50e7

100

-19

100

%


3.87e6


d

Impurities

%

2.00 4.00 6.00 8.00 10.00 12.00 14.00 16.00 18.00 20.00 22.00 24.00Time-20

100

%

1.50e7Degradants

2.00 4.00 6.00 8.00 10.00 12.00 14.00 16.00 18.00 20.00 22.00 24.00Time-20

Composite sample

A threshold setting for each sub-sample is important to pick up only the components of interest.

In case of a synthesis batch, the threshold value should be low to label all ypeaks above 0.05%.

In case of forced degradation samples, the threshold could be high to match only primary degradation compounds (in this example, a threshold setting of 5 0% is used)



5.0% is used).

Trial & Error

Method Development by Trial-and-ErrorGenerally one single type of column is chosen at randomOth h t hi t l t d t Other chromatographic parameters are selected at random and adapted in several experiments (composition mobile phases, pH, buffer, column temperature, gradient)

Con’sStep by step optimization without scientific background resuslting in a method which by definition is not the g yoptimum separationVery Time ConsumingManual Raw Data evaluationNo Peak Tracking



OVAT

4.5

5

3

3.5

4 0.75

0.80

0.95

1.151.151.25 1.55

2.15

2

2.5

3

% B

/min 0.85

1

1.5

0.90

1.05

0.751.251.501.00

0

0.5

0 5 10 15 20 25 30 35 40 45 50 55 60

0.65

0.80



Temp

DoE

6

4

5

3

% B

/min

2

0

1

0 5 10 15 20 25 30 35 40 45 50 55 60 65 70



0 5 10 15 20 25 30 35 40 45 50 55 60 65 70

Temp

Cl i l h d d l h h T i l & E ill

RoadblocksClassical method development approaches such as Trial & Error are still common practice. These step-by-step optimizations (One Variable at a Time) are very time-consuming and result in methods which are, by definition, not the optimum separations.

The major roadblock with UV-based detection approaches is the tracking of the peaks between all performed experiments.

mAU

75

100

125

150

175

min6 8 10 12 14 16 18 20

0

25

50

mAU

175

25

50

75

100

125

150



min6 8 10 12 14 16 18 20

0

COSMOST id ll th di d t th COSMOS d l d

This innovative Screening and Method Development system is based on 3 l

To avoid all these disadvantages, the COSMOS process was developed.

Computer Organized Screening and Method Optimization System

principlesthe use of a scientific strategy (DoE)the introduction of a single quadrupole mass spectrometer beside the PDA detector (LC-MS)

mAU

125

150

175

mAU

125

150

175

mAU

125

150

175

a special designed software for automated peak tracking and data evaluation

min6 8 10 12 14 16 18 20

0

25

50

75

100

mAUmin6 8 10 12 14 16 18 20

0

25

50

75

100

mAUmin6 8 10 12 14 16 18 20

0

25

50

75

100

mAUmAU

50

75

100

125

150

175

mAU

50

75

100

125

150

175

mAU

50

75

100

125

150

175



min6 8 10 12 14 16 18 20

0

25

50

min6 8 10 12 14 16 18 20

0

25

50

min6 8 10 12 14 16 18 20

0

25

50

COSMOS Strategy

Screening a matrix of all chromatographic parameters (full factorial DoE) takes too long

e.g. For 5 separation parameters (column, pH, organic modifier, temperature, and gradient)= (5 x 4 x 7 x 2) x (35 + 65) min = 28 000 min

± 467 h ± 20 d f b l = ± 467 hours or ± 20 days for one sub-sample

THEREFORE …

The most important chromatographic parameters are optimized in multiple WAVES based on a scientific approach (DoE).pp ( )

Chromatographic columns and pH are two of the most important parameters contributing to selectivity on reversed-phase HPLC. A wide pH range on each column is therefore screened first.

WAVE 1: Column screening and pH optimization5 x 4 = 20 x 35 min = 700 min = 11h40Screenings-module implemented in 1997 (without MS)

WAVE 2 O i difi i i iWAVE 2: Organic modifier optimization7 x 35 min = 245 min = 4h05

WAVE 3: Temperature and Gradient optimization(2 x 35 min) + (2 x 65 min) = 200 min = 3h20



TOTAL = ± 19 hours for one sub-sample

Wave 1: Column - pH ScreeningpH 2.5 pH 4.8

Column 1 XBridge C18 - 3.5 μm (150 x 4.6 mm) Column 1 XBridge C18 - 3.5 μm (150 x 4.6 mm)Column 2 XBridge Shield RP18 - 3.5 μm (150 x 4.6 mm) Column 2 XBridge Shield RP18 - 3.5 μm (150 x 4.6 mm)Column 3 XBridge Phenyl - 3.5 μm (150 x 4.6 mm) Column 3 XBridge Phenyl - 3.5 μm (150 x 4.6 mm)Column 4 Ascentis Express C18 - 2.7µm (150 x 4.6mm) Column 4 Ascentis Express C18 - 2.7µm (150 x 4.6mm)Column 5 Atlantis T3 - 3 μm (150 x 4.6 mm) Column 5 Atlantis T3 - 3 μm (150 x 4.6 mm)

Solvent A 5 mM NH4TFA + 10 mM NH4Ac in water + 0.1% TFA Solvent A 5 mM NH4TFA + 10 mM NH4Ac in water + 0.05% HAc

pH 2.5 pH 4.8

Solvent B CH3CN + 0.1% TFA Solvent B CH3CN + 0.05% HAcGradient Gradient

Time 0 20 25 27 35 Time 0 20 25 27 35% A 95 0 0 95 95 % A 95 0 0 95 95% B 5 100 100 5 5 % B 5 100 100 5 5

Flow 1 ml/min Flow 1 ml/minColumn temperature 45°C Column temperature 45°C

Column 1 XBridge C18 - 3.5 μm (150 x 4.6 mm) Column 1 XBridge C18 - 3.5 μm (150 x 4.6 mm)Column 2 XBridge Shield RP18 - 3.5 μm (150 x 4.6 mm) Column 2 XBridge Shield RP18 - 3.5 μm (150 x 4.6 mm)Column 3 XBridge Phenyl - 3.5 μm (150 x 4.6 mm) Column 3 XBridge Phenyl - 3.5 μm (150 x 4.6 mm)Column 4 Ascentis Express C18 - 2.7µm (150 x 4.6mm) Column 4 Ascentis Express C18 - 2.7µm (150 x 4.6mm)C l 5 Atl ti T3 3 (150 4 6 ) C l 5 Atl ti T3 3 (150 4 6 )

pH 6.9 pH 9.2

Column 5 Atlantis T3 - 3 μm (150 x 4.6 mm) Column 5 Atlantis T3 - 3 μm (150 x 4.6 mm)

Solvent A 5 mM NH4TFA + 10 mM NH4Ac in water Solvent A 5 mM NH4TFA + 10 mM NH4Ac in water + 0.025% NH4OHSolvent B CH3CN Solvent B CH3CN + 0.025% NH4OHGradient Gradient

Time 0 20 25 27 35 Time 0 20 25 27 35% A 95 0 0 95 95 % A 95 0 0 95 95% B 5 100 100 5 5 % B 5 100 100 5 5

20 experiments for each sub-sample are conducted. When all of the data are collected foreach pH/column combination, the peaks are matched. A resolution map is generated for

h l i i di ti f th ti l H f h Th h t t d

Flow 1 ml/min Flow 1 ml/minColumn temperature 45°C Column temperature 45°C



each column, giving a prediction for the optimal pH for each. The chromatograms are rated,and the optimal column and pH is chosen.

Column

312 3

45 6

79

10

8 11

3 9Platinum EPS C18XBridge C18

4

3

5

1

2 89

67

10

111



pH

pH 2.5

pH 7



Wave 2: Organic modifier Screening

Column XBridge C18 - 3.5 μm (150 x 4.6 mm)

C1CH3CN

C1CH3CN

Solvent A 5 mM NH4TFA + 10 mM NH4Ac in water + 0.1% TFA

Solvent B OM1 CH3CN + 0.1% TFAOM2 CH3OH + 0.1% TFAOM3 CH3CN - IPA (50/50) + 0.1% TFAOM4 CH3CN - CH3OH (50/50) + 0.1% TFA

C2 C3

C4 C5

C6

CH3CN - CH3OH (50/50)

CH3CN - IPA (75/25)

C2 C3

C4 C5

C6

CH3CN - CH3OH (50/50)

CH3CN - IPA (75/25)

O C 3C C 3O ( / ) . %OM5 CH3CN - IPA (75/25) + 0.1% TFAOM6 CH3OH - CH3CN - IPA (50/25/25) + 0.1% TFA

GradientTime 0 20 25 27 35% A 95 0 0 95 95

C2 C3C6CH3OH CH3CN - IPA

(50/50)CH3OH - CH3CN - IPA

(50/25/25)

C2 C3C6CH3OH CH3CN - IPA

(50/50)CH3OH - CH3CN - IPA

(50/25/25)

% B 5 100 100 5 5

Flow 1 ml/min

Column temperature 45°C

On the best column and with the optimized aqueous mobile phase from Wave 1, sevenexperiments (derived from the Snyder’s solvent triagle) with different organic modifiers areperformed to predict the optimal organic mobile phase composition



performed to predict the optimal organic mobile phase composition.

Organic modifier

CH3CN

CH3CN - MeOH

CH3CN - IPA



Wave 3: Gradient - Temperature ScreeningColumn XBridge C18 - 3.5 μm (150 x 4.6 mm) Column XBridge C18 - 3.5 μm (150 x 4.6 mm)

Solvent A 5 mM NH4TFA + 10 mM NH4AC + 0.1% TFA Solvent A 5 mM NH4TFA + 10 mM NH4AC + 0.1% TFASolvent B CH3CN - MeOH (50/50, v/v) Solvent B CH3CN - MeOH (50/50, v/v)

Gradient GradientTime 0 20 25 27 35 Time 0 50 55 57 65% A 95 0 0 95 95 % A 95 0 0 95 95% B 5 100 100 5 5 % B 5 100 100 5 5

Flow 1 ml/min Flow 1 ml/min

Column temperature 30°C Column temperature 30°C

Column XBridge C18 - 3.5 μm (150 x 4.6 mm) Column XBridge C18 - 3.5 μm (150 x 4.6 mm)

Solvent A 5 mM NH4TFA + 10 mM NH4AC + 0.1% TFA Solvent A 5 mM NH4TFA + 10 mM NH4AC + 0.1% TFASolvent B CH3CN - MeOH (50/50, v/v) Solvent B CH3CN - MeOH (50/50, v/v)

Gradient GradientTime 0 20 25 27 35 Time 0 50 55 57 65% A 95 0 0 95 95 % A 95 0 0 95 95% B 5 100 100 5 5 % B 5 100 100 5 5

Flow 1 ml/min Flow 1 ml/min

On the best column and with the optimized aqueous mobile phase from Wave 1, and withthe optimized organic modifier out of Wave 2, four experiments are conducted to predict

Column temperature 60°C Column temperature 60°C



p g p pthe optimal gradient slope and column temperature.

Gradient Slope

25 min 30 min

35 min

10 mM NH4OAC + 0.01% HAc 95 30CH3CN - MeOH (60/40) 5 70



Temperature

RT

45°C



COSMOS Equipment (Twins)



Manual process for Data evaluation

Make a stacked overlay of PDA (MaxPlot) and MS (TIC) signal

Define the correlating mass for each PDA peak or vice versa

Generate a component list (Excel® Component Table) of all masses found across all experiments with corresponding RT

Reject false positives (Blank peaks, Ghost-peaks, baseline j p ( p , p ,fluctuations, …)

All labeled components are listed into one Component Table which can be transferred to commercial chromatographic experimental g p pdesign software.



Stacked Overlay PDA - TIC

W1_COL4_25_S1

100W 1_COL4_25_S1 2: Diode Array

TIC6.71e7

8.00

9.27

% 6.37

5.40

4 66

6.63

7 82

8.16

8.33

11.06

10.90

11.90

100

-10

4.66 7.82 8.92

W 1_COL4_25_S1 1: Scan ES+ TIC

5.50e711.03

8.93 10.90

%

8.90

8.188.006.64

6.38

4.66

4.825 41

7.83

8.339.26 11.89

4.00 5.00 6.00 7.00 8.00 9.00 10.00 11.00 12.00 13.00Time-10

5.41



Define correlating MH+ and RTW1_COL4_25_S1

100W1_COL4_25_S1 396 (6.627) 1: Scan ES+

2.10e7411

MH+ 411W1_COL4_25_S1


TIC6.71e7

27

%

412

10

%

RT

6.63

RT

9.2

150 175 200 225 250 275 300 325 350 375 400 425 450 475 500 525 550 575 600 625 650 675 700 725 750 775 800 825 850 875 900m/z0

206159167 192 227 368348328295230 259 280 326 370 402

413

415 844454 784584564522507490486 561 742656646607622 665687 721697 771 812 835 863 886 898

W1_COL4_25_S1

100W1_COL4_25_S1 554 (9.265) Cm (552:559) 1: Scan ES+

1.35e7314

MH+ 314

100

%

-10W1_COL4_25_S1 1: Scan ES+

TIC5.50e7

%4.00 5.00 6.00 7.00 8.00 9.00 10.00 11.00 12.00 13.00 14.00

Time-10

Column 4 pH 2.5

150 175 200 225 250 275 300 325 350 375 400 425 450 475 500 525 550 575 600 625 650 675 700 725 750 775 800 825 850 875 900m/z0

297234205172151 195 242 286276

315

627

316355

328 408402364 472459417 601523507500 563547 575 613628

649640 662 785716680 722 764 796 840817 880876 888897

Generate a list of all masses found with June 9 - 10, 2005ANG annual meeting

21Rudy Sneyers J&J PRDAnalytical Development - DS&CC 21

21

Generate a list of all masses found with corresponding RT

Co-elution

W1_COL4_25_S1

100W1_COL4_25_S1 478 (7.996) Cm (476:483) 1: Scan ES+

9.55e6396

MH+ 380 - 396 - 426

W1_COL4_25_S1


TIC6.71e7

00

380

10

% RT

8.%

426

397

100

%

-10W1_COL4_25_S1 1: Scan ES+

TIC5.50e7

150 175 200 225 250 275 300 325 350 375 400 425 450 475 500 525 550 575 600 625 650 675 700 725 750 775 800 825 850 875 900m/z0

245194182151 239230 348246 346328276 306 378

381

398

402

428

429

760430 531448477 523482 533547 587 755606 654607 651 715684705 776 782 821830854 876877 8974.00 5.00 6.00 7.00 8.00 9.00 10.00 11.00 12.00 13.00

Time-10

Column 4 pH 2.5

In case of co-elution, it can be expected that within 20 experiments all peaks are likely to be resolved at least once and the corresponding masses added to the Component Table



Component Table

Co-elution

W1_COL1_25_S1

100W1_COL1_25_S1 529 (8.861) 1: Scan ES+

2.03e7426

MH+426

W 1 _ C O L 1 _ 2 5 _ S 1

1 0 0

%

W 1 _ C O L 1 _ 2 5 _ S 1 2 : D io d e A r ra y T IC

6 .5 3 e 7

RT

8.81

%

428

1 0 0

%

-1 0W 1 _ C O L 1 _ 2 5 _ S 1 1 : S c a n E S +

T IC9 .7 2 e 7

150 175 200 225 250 275 300 325 350 375 400 425 450 475 500 525 550 575 600 625 650 675 700 725 750 775 800 825 850 875 900m/z0

155348259235195192 205 328300279 411402

394370

429

430 448 707657561481 521486 547 603599 619622 673 683 873709 743 853757 819806759 841 877891

W1_COL3_90_S1

100W1_COL3_90_S1 745 (12.505) 1: Scan ES+

9.20e7380

5 .0 0 6 .0 0 7 .0 0 8 .0 0 9 .0 0 1 0 .0 0 1 1 .0 0 1 2 .0 0 1 3 .0 0 1 4 .0 0T im e-1 0

Column 1 pH 2.5W 1 _ C O L 3 _ 9 0 _ S 1

1 0 0W 1 _ C O L 3 _ 9 0 _ S 1 2 : D io d e A r ra y

T IC 100

%

9.20e7

MH+ 380

-1 0

1 0 0

%

T IC5 .4 3 e 7

W 1 C O L 3 9 0 S 1 1 : S c a n E S +

RT

12.4

8

%

194151 165378

245205 222 304300256 327 344

381

382 443383 402

411

822759444

721488465 504 532 644545 578 663 683 703 723 761781 798 824844883854 885

1 0 0

%

W 1 _ C O L 3 _ 9 0 _ S 1 1 : S c a n E S + T IC

3 .2 0 e 8



150 175 200 225 250 275 300 325 350 375 400 425 450 475 500 525 550 575 600 625 650 675 700 725 750 775 800 825 850 875 900m/z0

194165 245 304300 327 344 411 721488 504 532 644578 663 683 703 723 761 883854 885

7 .0 0 8 .0 0 9 .0 0 1 0 .0 0 1 1 .0 0 1 2 .0 0 1 3 .0 0 1 4 .0 0 1 5 .0 0 1 6 .0 0 1 7 .0 0 1 8 .0 0 1 9 .0 0 2 0 .0 0 2 1 .0 0 2 2 .0 0 2 3 .0 0T im e-1 0

Column 3 pH 9.0

Excel® Component Table

MassW1_COL1_10_25

W1_COL1_10_48

W1_COL1_10_70

W1_COL1_10_90

W1_COL2_10_25

W1_COL2_10_48

W1_COL2_10_70

W1_COL2_10_90

W1_COL3_10_25

W1_COL3_10_48

W1_COL3_10_70

W1_COL3_10_90

RT RT RT RT RT RT RT RT RT RT RT RT205 6.11 5.44 7.67 9.56 4.97 5.04 7.56 9.55 5.79 6.14 9.08 9.89245 9 12 12 3 12 3 12 3 7 76 11 99 12 13 12 04 8 66 11 78 11 94 11 93245 9.12 12.3 12.3 12.3 7.76 11.99 12.13 12.04 8.66 11.78 11.94 11.93297 11.05 12.84 14.47 14.57 9.71 12.13 14.02 10.01 10.41 12.33 13.88 13.84300 11.46 11.73 11.7 11.68 10.33 11.44 11.53 11.41 11.22 11.41 11.55 11.52309 6.47 8.48 9.75 9.8 5.66 8.62 10.01 10.1 6.28 8.78 10.11 10.1314 9.41 10.9 10.84 10.8 8.66 11.32 11.4 11.29 9.44 10.71 10.81 10.74340 9.54 9.57 9.57 9.2 7.89 9.27 9.47 8.94 9.34 9.28 9.39 8.98368 7 04 6 62 7 4 9 13 6 8 6 85 7 78 9 63 7 19 7 21 9 2 9 52368 7.04 6.62 7.4 9.13 6.8 6.85 7.78 9.63 7.19 7.21 9.2 9.52369 13.07 14 19.16 19.72 11.59 13.35 18.09 18.53 12.53 14.36 18.25 17.49380 9.25 9.09 12 12.59 8.42 8.95 12.03 12.54 9.02 9.71 12.52 12.4396 9.2 8.94 11.5 12.15 8.32 8.9 11.86 12.49 9.02 9.64 12.35 12.05411 7.89 8.55 9.97 11.48 7.09 8.45 9.96 11.52 7.59 9.31 12.15 11.7415 13.61 16.29 17.43 17.54 11.94 15.32 16.74 16.63 12.68 15.14 16.23 16.14416 9.24 10.83 14.54 15.27 8.18 10.6 14.29 14.91 8.94 11.46 14.79 14.44426 8.78 8.48 10.26 11.04 8.41 8.62 10.63 11.46 8.8 9.04 11.31 11.22459 9.79 11.1 12.97 14.7 8.74 10.88 13.05 14.67 9.49 12.15 16.39 15.09531 9.56 12.47 13.49 13.56 8.85 12.04 13.25 13.24 9.14 12.47 13.6 13.57705 12.92 16.54 16.47 16.51 11.72 15.99 16.15 15.88 12.4 15.98 16.05 15.45

X 17.67 17.67 17.75 17.71 15.43 17.28 17.66 17.37 17.44 15.45 15.64 15.59

The Component Table can be transferred to advanced experimental designsoftware (LC Simulator) where the experiments are modelled and a resolution map is generated, giving a prediction for the optimum for each examined chromatographic variable



chromatographic variable.

COSMOS Software

Rigorous method development performed with composite samples and g p p p pmultiple chromatographic detectors results in a tremendous amount of hyphenated data files. This amount of data makes it cumbersome to see the whole story that the data is telling.One way to organize these data is the collection of the relevant peak data in One way to organize these data is the collection of the relevant peak data in one single location. The Method Development Console (AutoChrom) is designed to summarize these data while retaining full links back to the hyphenated raw data. After data acquisition the detector signals are directly imported to an After data acquisition, the detector signals are directly imported to an AutoChrom workspace, then UV and MS Peak matching are applied to label the peaks in all experiments. Compounds with either UV or MS response are being picked up.All l b l d t li t d t ti ll i t C t T bl All labeled components are listed automatically into one Component Table which can be passed to LC Simulator for modeling and prediction.



AutoChrom Method Development Cycle

Acquire data

Automatically track peaks between injections

min6 8 10 12 14 16 18 20

mAU

0

25

50

75

100

125

150

175

mAU

50

75

100

125

150

175

min6 8 10 12 14 16 18 20

mAU

0

25

50

75

100

125

150

175

mAU

50

75

100

125

150

175

min6 8 10 12 14 16 18 20

mAU

0

25

50

75

100

125

150

175

mAU

50

75

100

125

150

175

between injections

min6 8 10 12 14 16 18 20

0

25

min6 8 10 12 14 16 18 20

0

25

min6 8 10 12 14 16 18 20

0

25

Separation acceptable?

Create peak table

YES

Final Method



p


Acquire data


min6 8 10 12 14 16 18 20

mAU

0

25

50

75

100

125

150

175

mAU

50

75

100

125

150

175

min6 8 10 12 14 16 18 20

mAU

0

25

50

75

100

125

150

175

mAU

50

75

100

125

150

175

min6 8 10 12 14 16 18 20

mAU

0

25

50

75

100

125

150

175

mAU

50

75

100

125

150

175

between injectionsSelect next experiment

min6 8 10 12 14 16 18 20

0

25

min6 8 10 12 14 16 18 20

0

25

min6 8 10 12 14 16 18 20

0

25


Create peak table

NOModel separations

Create resolution map



p


Acquire data


min6 8 10 12 14 16 18 20

mAU

0

25

50

75

100

125

150

175

mAU

50

75

100

125

150

175

min6 8 10 12 14 16 18 20

mAU

0

25

50

75

100

125

150

175

mAU

50

75

100

125

150

175

min6 8 10 12 14 16 18 20

mAU

0

25

50

75

100

125

150

175

mAU

50

75

100

125

150

175

between injections

min6 8 10 12 14 16 18 20

0

25

min6 8 10 12 14 16 18 20

0

25

min6 8 10 12 14 16 18 20

0

25


Create peak table

YES

Final Method



p

AutoChrom Workspace

The AutoChrom workspace is designed to organize the different waves. In the left example, Wave 1 is setup for five columns four pHs three five columns, four pHs, three sub-samples, and two detectors. Once a wave is created with the AutoChrom wizard, it can be saved as a template and will be available for the next application.

Wave 1 Wave 2 Wave 3 All Waves

Some example of method development workflows showing individual optimization and screening

Wave 1 Wave 2 Wave 3 All Waves

optimization and screening waves. The project can link and organize instrument data by wave, linking organization of the data to the decision-

ki



making process.

MS Signals



W1_COL2_25_S1 JNJ27387581_RSV 08-Mar-2006

IntelliXtract : baseline correction & peak tracking

15:12:07


2303.20e5

%

100

-16W1_COL2_25_S1 1: Scan ES+

TIC5.88e7

%

2.00 4.00 6.00 8.00 10.00 12.00 14.00 16.00 18.00Time-10



IntelliXtract done!



MS Peak Matching done



MS Peak Matching 100% accurate

UV Signals



UV Peak Matching done

UV Peak Matching ± 60% accurateJune 9 - 10, 2005ANG annual meeting

36Rudy Sneyers J&J PRDAnalytical Development - DS&CC 36

36

g

Reconciliation



Composite Peak Table

After the UV- and MS-MAP algorithms are executed, the results (tables of masses andcorresponding retention times) are directly imported into the AutoChrom workspace. After



completion of all sub-samples from the entire workflow, a component table can begenerated.

Simulated Composite chromatogram after reconciliation



LC Simulator



Wave 1: Resolution Map

A quadratic polynomial model based on the 4 measurements is constructed in order to predictretention times as a function of the pH. The predicted retention times at intermediate pH-valueswith their corresponding peak widths (considered constant since gradient elution) are sorted fromhigh to low, and finally the resolutions between successive peaks are calculated and the Rsmin can



g , y p minbe predicted for each pH value and the optimal pH can be determined.


(50)Min Rs 1.62

50% ACN - 40% MeOH - 10% X

(55)Min Rs 1.41

60% ACN - 30% MeOH - 10% X% % %

(60)Min Rs 1 28

Two strategies were applied to predict from the seven experiments, derived from a Snyder’s solvent triangle, at which mobile phase composition, i.e., at which fractions of organic modifiers,

Min Rs 1.2870% ACN - 30% MeOH

the highest Rsmin values (i.e., the best separation for the worst separated peak pairs) can be reached. The first one explores the three sides of the solvent triangle one by one, whilst the second method investigates the complete triangle. Both strategies first predict the retention times and peak widths of each compound at every considered mobile phase composition by using a mathematical model, then retention times with their corresponding widths are sorted from high to low, and finally the



p g g yresolutions between successive peaks are calculated. In that way, the Rsmin can be predicted for each mobile phase composition and the optimal mobile phase can be determined.


This wave is a 2-D optimization mode resulting in a 3-D resolution map. Small changes of theinvestigated parameters can influence the deterioration of the minimal resolution. The areas withgood resolution are colored in red. A large area implicates a robust separation.Method suitability is a function defining the quality of the proposed separation in terms of runtime robustness resolution of the closest eluting peaks and minimum retention factor The optimal



time, robustness, resolution of the closest eluting peaks, and minimum retention factor. The optimalconditions with regard to resolution and suitability can be shown on the map (•).

R213129 - GLYT1Theoretical vs Practical



COSMOS second generation

By combining high resolution and speed of the UPLC chromatographic separation together with advanced MS chromatographic separation together with advanced MS detectors and software tools, it significantly improves the efficiency of method development.Due to the analysis time reduction with UPLC it is ypossible to speedup method development or cover more space of the DoE by combining wave 1 and 2.

(1 UPLC-SQD vs 3 HPLC-ZQ’s)(1 UPLC-SQD vs 3 HPLC-ZQ s).

On each column 4 pH’s and 3 organic modifiers are screened.

4 columns x 4 pH’s x 3 OM = 48 experimentsIn a separated wave 1 and 2 only 22 experiments are performed



performed

Combined wave 1 and 2

Column 1 Acquity BEH C18 - 1.7 µm (150 x 2.1 mm) Column 1 Acquity BEH C18 - 1.7 µm (150 x 2.1 mm)Column 2 Acquity BEH Shield RP - 1.7 µm (150 x 2.1 mm) Column 2 Acquity BEH Shield RP - 1.7 µm (150 x 2.1 mm)Column 3 Acquity BEH Phenyl - 1.7 µm (150 x 2.1 mm) Column 3 Acquity BEH Phenyl - 1.7 µm (150 x 2.1 mm)Column 4 Acquity HSS T3 - 1.8 µm (150 x 2.1 mm) Column 4 Acquity HSS T3 - 1.8 µm (150 x 2.1 mm)

Solvent A 5 M NH TFA + 10 M NH AC i t + 0 1% TFA Solvent A 5 M NH TFA + 10 M NH AC i t + 0 05% HA

pH 4.8pH 2.5

Solvent A 5 mM NH4TFA + 10 mM NH4AC in water + 0.1% TFA Solvent A 5 mM NH4TFA + 10 mM NH4AC in water + 0.05% HAcSolvent B CH3CN Solvent B CH3CNSolvent C CH3OH Solvent C CH3OHSolvent D CH3CN - IPA (50/50) Solvent D CH3CN - IPA (50/50)

Gradient GradientTime 0 10 12 12.5 15 Time 0 10 12 12.5 15% A 95 0 0 95 95 % A 95 0 0 95 95% B 5 100 100 5 5 % B 5 100 100 5 5

Flow 0.4 ml/min Flow 0.4 ml/minColumn temperature 45°C Column temperature 45°C

pH 9pH 7

Column 1 Acquity BEH C18 - 1.7 µm (150 x 2.1 mm) Column 1 Acquity BEH C18 - 1.7 µm (150 x 2.1 mm)Column 2 Acquity BEH Shield RP - 1.7 µm (150 x 2.1 mm) Column 2 Acquity BEH Shield RP - 1.7 µm (150 x 2.1 mm)Column 3 Acquity BEH Phenyl - 1.7 µm (150 x 2.1 mm) Column 3 Acquity BEH Phenyl - 1.7 µm (150 x 2.1 mm)Column 4 Acquity HSS T3 - 1.8 µm (150 x 2.1 mm) Column 4 Acquity HSS T3 - 1.8 µm (150 x 2.1 mm)

Solvent A 5 mM NH4TFA + 10 mM NH4AC in water Solvent A 5 mM NH4TFA + 10 mM NH4AC in water + 0.025% NH4OHSolvent B CH3CN Solvent B CH3CNSolvent C CH3OH Solvent C CH3OH3 3

Solvent D CH3CN - IPA (50/50) Solvent D CH3CN - IPA (50/50)

Gradient GradientTime 0 10 12 12.5 15 Time 0 10 12 12.5 15% A 95 0 0 95 95 % A 95 0 0 95 95% B 5 100 100 5 5 % B 5 100 100 5 5

Flow 0.4 ml/min Flow 0.4 ml/min



0. / 0. /Column temperature 45°C Column temperature 45°C

COSMOS12 Waters XBridge C18; 150mmx4.6mm;3.5µmRSNE

15:27:5821-Mar-2007Cosmos Mix

COSMOS with UPLC

1 5e+1

2.0e+1

col1_25_OM1_S1 2: Diode Array Range: 2.611e+110

AU

5.0

1.0e+1

1.5e+1

12

3

4

57

126

8

91113

1415

16

4 5e+1

2.00 3.00 4.00 5.00 6.00 7.00 8.00 9.00 10.00 11.00 12.00 13.00 14.00

0.0

W12_COL1_25_OM1_S1 2: Diode Array Range: 4.8e+1

10

AU 2.5e+1

3.0e+1

3.5e+1

4.0e+1

4.5e+1

3

10

6 8A

5.0

1.0e+1

1.5e+1

2.0e+1

12

3

4 57

12

6+8

911 13

1415

16



Time2.00 3.00 4.00 5.00 6.00 7.00 8.00 9.00 10.00 11.00 12.00 13.00 14.00

0.0

Method Development Suite (ACD/Labs)

Set upInstrument Methods

and Sample Sets

Set up AutoChrom

and Sample Sets for Wave 1, 2 and 3 in

CDS system

pWorkspace for Wave 1 – 2 - 3

Transfer data with A t ti S

G t C t

Automation Server

Run IntelliXtract

Generate Component Table with

Processing Wizard 1 and 2



AutoChrom for Empower (ACD/Labs)

Set up AutoChrom Workspace for

Wave xx (strategy)

Instrument Methods and Injections are

generated by AutoChromAutoChrom

INSTRUMENT CONTROL

Data are directly transfered in AutoChrom

Run IntelliXtract

Generate Component Table with

Processing Wizard xx



xx

Process Output & Return

30

LC method development time and cost

20

25

10

15

Tim

e (h

)

0

5

1 2 3 4 5Dw ell time in days Cost in k$

1. Trial & error 2. Screenings-module 3. MS Twin approach 4 COSMOS (HPLC) 5 COSMOS (UPLC)



4. COSMOS (HPLC) 5. COSMOS (UPLC)

New AutoChrom Tools

Safety netWith this new AutoChrom tool it is possible to detect new impurities.p pAfter running e.g. Wave 1 on a new production batch the software is capable to asign automatically new components by comparing the generated component table with the table of the last developed methodmethod.

Import of component listFor method development of known components (standards p p (available) it is possible to import a list of MW’s or structures into the software. Only these components will be tracked and summarized in the component table.

How to handle isomers (Under investigation)If standards are available bring the isomers in different sub-samples. The same can be done for compounds without MS



response.

Conclusions

Select the most appropriate samples from forced degradation studies and the representative synthesis route to develop purity and stability indicating methods in order to demonstrate the specificity and suitability of chromatographic methods.The use of mass spectrometry with dedicated evaluation and The use of mass spectrometry with dedicated evaluation and modeling software is an innovative way to perform method development based on scienctific principles.COSMOS will significantly improve quality (reliability, robustness, g y p q y ( y, ,and life time) of the methods, reduces development time and costsbut also reduces loss of interpretation information and expensive retesting of samples. Wi h h h h d d f d i d h With the hyphenated data from modern instruments and the available computer capacity, AutoChrom is a tool for organizing, visualizing, and tracking the data retaining full links back to the original raw data.



original raw data.

Acknowledgement and Contact Information

The authors acknowledge ACD/Labs and the University of Brussels (VUB) for the collaboration and their contribution to this research project.

Mike McBrien and Mark BaylissAdvanced Chemistry Development, Inc. (ACD/Labs)Toronto, Ontario

M. Dumarey, A.M. van Nederkassel, Y. Vander HeydenVrije Universiteit Brussels (VUB)Vrije Universiteit Brussels (VUB)Department of Analytical Chemistry and Pharmaceutical TechnologyJette, Belgium

CONTACT INFORMATIONRudy Sneyers

rsneyers@prdbe jnj [email protected]℡ +32.14.60.29.83

Willy Janssenswjansse2@prdbe jnj com



[email protected]℡ +32.14.60.37.79

Backup



Process COSMOS Software

6. 5.

RECONCILIATION4.

COMPONENT UV & MS TABLE

6. MODELING

7.

3. UV & MS

PEAK MATCHING

RESOLUTIONMAP

9

2. DATA TRANSFER

8. SELECT BEST

1. ACQUISITION

9. NEXT WAVE

Automation Server


Analytical Development - DS&CC 55Chromatographic Society Spring Symposium 19 – 20th May 2010 - Hoddesdon, UK

Threshold Settings


Analytical Development - DS&CC 56Chromatographic Society Spring Symposium 19 – 20th May 2010 - Hoddesdon, UK 56

Orthogonal Methods

16

20

0.980738

16

18

20

0.590822

8

12

C325

8

10

12

14

C42

5

44 6 8 10 12 14 16

C125

6

8

4 6 8 10 12 14 16

C125

After the final method is established, all previously performed experiments can be ranked interms of orthogonality using the correlation coefficient between the retention times of a givenindividual experiment and the final method.

The orthogonal methods are applied to check the validity of the method in terms of specificityg pp y p yeach time

When a change is made in the synthetic routeWhen a change is made in suppliers of raw materials and excipientsWhen there is a change in formulation



g

Running Wave 1 is a better approach to perform orthogonal experiments (20 dimensions)

ChromGenius

ACD/ChromGenius can be trained with the previous identified structures. Simply enter experimental t ti ti d k idth i t Ch G i d t b ith th t tretention times and peak widths into a ChromGenius database with the structures.

This knowledge base of the structures can be used as a basis to predict retention times and chromatograms for new compounds based on chemical structure. Predictions are powered by physicochemical prediction software. Simply draw the structure of the new compound to be separated, and ACD/ChromGenius simulates the separation for each generic method In just a few seconds the methods are ranked according



simulates the separation for each generic method. In just a few seconds, the methods are ranked according to suitability and displayed with a simulated chromatogram, and a table of predicted retention times, without performing any injection

COSMOS with UPLC

10

21

3

134 12

515

16

149

7 6

11

8

10

1611

10

3

12

13

15

16

14

11+8

4 5 79

612



13 16

LC Simulator - Suitability Options



COSMOS second generation

Improvement of pH predictionp p pDue to the use of different additives to prepare mobile phases with different pH’s, additional selectivity differences are introduced which results in inaccurate predictions of the pH.Use of a new buffer to screen the whole pH range.



Selectivity differences

COSMOS Waters XBridge C18; 150mmx4.6mm;3.5µmWJAN

15:08:1019-Mar-2007Sample1

4.0e+1

5.0e+1

W 1_COL1_294_S1 2: Diode Array Range: 6.417e+19+10

COSMOS Waters XBridge C18; 150mmx4.6mm;3.5µmWJAN

15:08:1019-Mar-2007Sample1

4.0e+1

5.0e+1

W 1_COL1_294_S1 2: Diode Array Range: 6.417e+19+10

NH4FORA

U

0.0

1.0e+1

2.0e+1

3.0e+1

12

3

45

67

8

1112

13

1614

15

AU

0.0

1.0e+1

2.0e+1

3.0e+1

12

3

45

67

8

1112

13

1614

15

3 5e+1

4.0e+1

4.5e+1

0.00 2.00 4.00 6.00 8.00 10.00 12.00 14.00 16.00 18.00 20.00 22.00 24.00

-1.0e+1

W 1_COL1_294_S1 2: Diode Array Range: 4.906e+1

9

3 5e+1

4.0e+1

4.5e+1

0.00 2.00 4.00 6.00 8.00 10.00 12.00 14.00 16.00 18.00 20.00 22.00 24.00

-1.0e+1

W 1_COL1_294_S1 2: Diode Array Range: 4.906e+1

9

NH4TFA

AU

1.0e+1

1.5e+1

2.0e+1

2.5e+1

3.0e+1

3.5e+1

12

144

5 76

16

3

1110

8

12

AU

1.0e+1

1.5e+1

2.0e+1

2.5e+1

3.0e+1

3.5e+1

12

144

5 76

16

3

1110

8

12

NH4TFA

The use of NH4TFA results in a flat baseline and narrow peak widths.Though the pH is exactly the same, the elution order in both chromatograms is

Time0.00 2.00 4.00 6.00 8.00 10.00 12.00 14.00 16.00 18.00 20.00 22.00 24.00

0.0

5.0 15131112

Time0.00 2.00 4.00 6.00 8.00 10.00 12.00 14.00 16.00 18.00 20.00 22.00 24.00

0.0

5.0 15131112



g p y gdifferent.

Importing Structures



Method Development Suite (ACD/Labs)




Set up AutoChrom

and Sample Sets for Wave 1 in CDS system

Set up AutoChrom


Set up AutoChrom


Set up AutoChrom Workspace Wave 1

Transfer data with Automation Server





Generate Component

Automation Server

Run IntelliXtract

Generate Component

Automation Server

Run IntelliXtract

Generate Component

Automation Server

Run IntelliXtract



pTable with

Processing Wizard 1

pTable with

Processing Wizard 2

pTable with

Processing Wizard 2

strategic and structured approaches in automated …strategic and structured approaches in automated...

Documents